def _get_title(self, root: ET.Element) -> str:
title: str = find_text_or_default(
root=root,
xpath=
"bibliographic-data/technical-data/invention-title[@lang='EN']",
default='')
return title
def _get_tags(self, root: ET.Element) -> List[str]:
def convert_tags(orig: str) -> List[str]:
"""
>>> convert_tags('C01C 03/16')
'C01C'
"""
return orig.split(' ')[:1]
clfs: str = base.find_text_or_default(root, 'PRI-IPC', '')
return convert_tags(clfs)
def _get_docid(self, root: ET.Element) -> str:
def convert_docid(orig: str) -> str:
"""
>>> converters('PATENT-US-GRT-1993-05176894')
'199305176894'
"""
return ''.join(orig.split('-')[-2:])
docid: str = base.find_text_or_default(root, 'DOCNO', '')
return convert_docid(docid)
def create_name_mapping(self) -> Dict[str, str]:
mpg: Dict[str, str] = dict()
for fpath in self.iter_query_files():
with open(fpath, 'r') as fin:
xml_body: str = self.converter.escape(fin.read())
root: ET.Element = ET.fromstring(xml_body)
topic_num: str = find_text_or_default(root, 'NUM', '')
doc_root: ET.Element = get_or_raise_exception(root.find('DOC'))
docid: str = self.converter._get_docid(doc_root)
if topic_num == '' or docid == '':
raise AssertionError
mpg[topic_num] = docid
return mpg
def _get_text(self, root: ET.Element) -> str:
text: str = base.find_text_or_default(root, 'SPEC', '')
return text
def _get_title(self, root: ET.Element) -> str:
title: str = base.find_text_or_default(root, 'TITLE', '')
return title